The system

Why options, and why sell them

I assume you are an experienced options trader, so I won’t give you a primer. Still, let me explain how I see options and why I chose to sell them.

Beyond their fascinating complexity, options are the most versatile instrument in finance. They can be used to speculate or to hedge, over hours or over years, and their volume has exploded. The US options market now moves billions of dollars every day and increasingly influences the prices of the underlying instruments themselves.

In my search for the Holy Grail I realised early that classic stocks or index day-trading is essentially a bet on direction. People who sell various systems claim that there are so-called “high-probability” setups, but anything that simply goes long or short is close to a coin flip after costs. And any setup based on indicators or market regimes will be researched by other market participants and, sooner or later, traded away. Even more so in the age of AI.

Options change the shape of the bet. Out-of-the-money options have an asymmetric probability profile: a 15-delta put has roughly an 85% chance of expiring worthless. Sell it, and you have high chances of keeping the premium. But theory says the 15% of cases where you lose should cost exactly as much as the 85% where you win. A zero-sum game, says the theory. What does practice say?

The volatility risk premium

In practice, option sellers have historically been paid for carrying that risk, because option prices tend to imply bigger moves than the market then delivers. Over the same four years as our backtest, the VIX (the market’s priced-in estimate of S&P 500 volatility over the next 30 days) was above the volatility the index actually realised over the following month on 87% of days: on average 18.8 against 15.1, a gap of almost four volatility points. Academic research finds the same over decades of data (Bakshi and Kapadia, 2003; Carr and Wu, 2009). The gap is known as the volatility risk premium. It is the price the market pays for insurance against large moves, and it exists because sellers occasionally take large losses.

I saw this for myself when I started selling options with a fairly crude system. The results were encouraging but never consistent, in the classic pattern of options tail risk: with a 70–80% win rate you win often, but the 20–30% of losers take back most of the winners’ profits. Still, a small margin remains, and that margin can be your edge.

The edge in selling options is being paid to carry risk for a period of time. Unlike most edges claimed for stock day-trading, it is well documented.

So the challenge isn’t finding winners: sell out-of-the-money options and there are plenty. The hard part is keeping the winners’ premium by managing the losers. The real question is whether we can tilt the odds by finding “higher-probability” winners and avoiding “higher-probability” losers.

Market regimes

My whole system is built around answering one question:

Can market regimes be used to make short-option trades more profitable?

You might think I just answered that question negatively for stocks. But stock and options markets are not the same. I don’t know a single person or institution that buys or shorts a stock without aiming to win. In options, plenty of buyers expect to lose: when they buy options for hedging, the loss is the cost of insurance. That makes a technical edge in selling options harder to trade away than one in stocks.

A short option also gives you a margin for error. With a stock, the price has to move your way from your entry for you to win. Sell an out-of-the-money put in a rising market, and you also win if the trend stops, or even reverses a little.

Let me explain where my research led. Having tested thousands of combinations, I found three distinct market regimes that influence the profitability of short-option trades:

1. Trending. The market is moving with momentum in one direction. You sell on the side it is moving away from: puts in a rising market, calls in a falling one.

2. Ranging. The market is going nowhere. You sell both sides: a strangle.

3. Reversal. A move has stretched too far, too fast. You sell against its continuation: after a drop, for example, a put spread below the market.

How do you decide whether a market is trending, ranging or reversing? Not by looking at a chart; I tried that a lot, and it doesn’t work. It takes specific, measurable criteria, and those are my secret sauce. They took countless hours to discover and I won’t say more than the obvious. AI was a big help. What matters is that these regimes can all appear in one day, or only one of them. They can last the whole day or only a few minutes, and quite often they overlap. That may seem counter-intuitive, but remember we are trading 0DTE: these are trends and ranges on minute time frames. Daily or longer charts don’t matter to us at all.

Why SPX, and why 0DTE

Before the specifics, let me explain why I tested and trade the system only on SPX, and only on same-day expiries.

SPX is the most liquid options market in the world, with a 0DTE expiry every weekday and ever-growing volume. As my system places a large number of trades every day, liquidity matters enormously: it drives the bid-ask spread, and through it, profitability. For the out-of-the-money strikes these strategies trade, the SPX 0DTE bid-ask spread is typically $0.05–$0.10 and widens only in exceptional moments. Over thousands of trades, every cent of slippage adds up.

SPX options are also European-style and cash-settled: no early assignment, and no shares delivered into your account at the close.

Could the system be run on SPY, probably the second most liquid options market in the world? In theory, yes. In practice, SPY is a tenth the size of SPX, so you need ten times as many contracts; commissions and slippage don’t shrink by anywhere near the same factor. They would eat almost all of the profit.

0DTE was obvious to me from the start. As an option seller I am after time decay, and it is at its fastest on expiry day. But the decisive point for me was ending every day flat. I see premium selling as an insurance business: I take on the risk of big moves and collect premium for it. Unlike a traditional insurer, though, I can close my position at any moment during trading hours, at a known cost. Every position is closed or settled by the end of the day, so nothing is held overnight. Even if there is after-hours trading, I need to sleep.

I have never understood the argument that 0DTE leaves no time to adjust a trade the way you can with 30 days to expiry. Adjusting a losing 30-day trade simply means setting up a new trade to try to recover the loss. That is no different from closing a 0DTE trade at a loss and opening a new one the next day.

The strategies

After years of research on 0DTE, I built a model that sells options every day, with entries governed by fixed rules that check the market regime: trending, ranging or reversing. Exits were tested just as extensively, and each strategy has its own stop-loss level (or, in one case, no stop at all), chosen to balance risk and return.

How the stops read: a stop at 100% closes the position when buying back the short leg would cost twice the credit received for it; at 150%, two and a half times. Range’s 400% stop closes a side when buying it back would cost five times the strangle’s combined credit.

The most important feature of the strategies, though, is that trades don’t happen at “trigger points”.

My model checks market conditions for each strategy every 15 minutes. It works like a traffic light per strategy. When the light is green, the model sells a put or a call (depending on the regime; Range sells both) at a pre-defined delta, and at the same time buys a further out-of-the-money option on the same side to form a spread.

My research has shown that long options trades, whatever the regime, consistently lose money over long periods. So without the long hedges I would make more money, in theory. I treat them as a cost of doing business, like the bid-ask spread and commissions, for two reasons. First, naked short options need far more margin than spreads: even a $500,000 account would reach its limit with a handful of contracts. Second, no backtest’s worst loss is the worst that can happen. In a catastrophic, unprecedented move, a spread’s loss is capped and known at entry; a naked option’s is not. Insurance companies re-insure their risks with specialised re-insurers as part of their cost of doing business. Buying the long leg is my re-insurance.

Once positions are open, they are managed by fixed exit rules, also backtested and optimised over thousands of trades. Three strategies have a stop-loss; Reversal strategy holds every position to settlement. No strategy takes profits early: every take-profit rule I tested made results worse. You will often hear elsewhere the advice to take profits at 50% of the credit. That rule comes from research on options with 30 to 45 days to expiry, and on minute-by-minute 0DTE data it did not hold up for my strategies.

Widths shown in the above table are for Portfolio Full. Portfolio Half uses 50-point spreads (25 for Reversal) and Portfolio Quarter 25-point spreads (15 for Reversal). Each strategy has its own page with a fuller description and its complete record.

Three ways to run it: Full, Half and Quarter

The sizing part of my system is not very flexible. The backtest of Full portfolio assumes every entry with 1 SPX contract. But it is many entries per day and since they are rules based, you need to execute them all as a set. Skip any trade and it is a different system. Entries tie up capital in the form of margin. Thankfully I have found a way to control margin requirement to some extent. That is with the width of the spreads.

The width of a spread sets how much margin it ties up. Its maximum loss is the width minus the credit received, times 100: a 100-point spread can lose up to $10,000, a 25-point spread up to $2,500. The system holds many spreads at once, so the margin adds up quickly. At Full size the Portfolio needed about $124,000 of margin on an average day and up to $443,000 on the busiest. Your account has to be sized for the busiest day, not the average one, because that is the day the margin is actually called on.

Not every trader can or wants to commit that much capital to one system. That is why Portfolio Half and Portfolio Quarter exist. They run exactly the same signals, on the same days, with the same deltas and stops; only the distance between the short strike and its hedge is narrower. That cuts the margin to roughly a half and a quarter of Full:

Portfolio, hypothetical / backtested, May 2022 – April 2026:

The names describe margin, and nothing else. Two things they do not describe. First, profit falls less than the capital does: Half keeps 77% of Full’s profit on about 49% of its margin, and Quarter keeps 47% on 24%. Measured against the account each tier needs, the annual return actually rises going down the tiers, from 40.8% at Full to 53.6% at Half and 60.7% at Quarter.

Second, and more important, risk does not shrink with a smaller peak margin; it grows, at least when each tier is run on an account sized to its own peak margin. Narrowing a spread cuts the credit you collect faster than it cuts what you can lose, so each trade keeps much of its capacity to lose while giving up more of its ability to win. Measured against each tier’s own peak margin, the worst drawdown goes from 12.5% at Full to 26.3% at Half and 49% at Quarter. Holding more capital than the peak margin lowers that percentage: run Quarter on a Full-sized account of $443,000, and its worst drawdown (−$52,750) is about 12% of it. But that is roughly the same as Full, for less than half of Full’s profit, because in dollars Quarter’s worst drawdown was almost as deep as Full’s (−$55,572). If you want to follow the system you must choose a tier. Do it based on how much drawdown you can live with, not only by the capital you have.

I started trading the system in May 2026. It was a good first month with ~25K $ of profit. Then came June 5th with a loss of more than 50K $ in one day. May cushion helped but I still landed with more than -25K $ in the red. This kind of loss happens rarely. But it can happen to you on day one or month two. You need to be able to absorb it. Since then I have earned this amount back many times over.

The portfolio

Each strategy has been backtested extensively and is profitable on its own, with a good risk-return profile. The real benefit, though, comes from running them together. They are built for different market regimes, and their results move largely independently. In the backtest, the daily P&L correlation between strategies ranged from −0.14 to +0.15, with one exception: the two Momentum strategies, at 0.47, which is no surprise since both trade with the trend. Yet all four can trade on the same day. So their returns add up while their risks partly cancel out.

The same effect shows up in margin. Momentum I or II may be selling puts while Reversal sells calls. A put spread and a call spread on the same expiry can’t both finish in the money, so the risk only has to be margined once (depending on your broker).

Here is what that adds up to (Portfolio Full, hypothetical / backtested, May 2022 – April 2026):

The profit is identical, because the trades are identical. But the Portfolio’s worst drawdown was less than half of the four drawdowns added together, and smaller even than Range’s alone (−$56,876). And because opposite-side positions share margin, the Portfolio needed about 20% less capital at its peak.

There is a second reason to trust a large sample over a good story: the law of large numbers. Any single 0DTE trade is close to noise. In the backtest, the average trade made just $78, against an average winner of $266 and an average loser of −$691, so a single trade tells you almost nothing. But the Portfolio placed 16,637 of them over 944 sessions, and over that many trades a small edge stops looking like luck. If the strategies had no real edge, the chance of luck alone producing an average day this far above zero over 944 sessions would be far smaller than one in a million. 39 of 48 months were profitable, there were never two losing months in a row, and every six-month stretch in the backtest made money; the worst one still made $59,805.

Statistics have a limit, though, and you should know it. They can tell you a result is unlikely to be chance, but not whether the rules were fitted to the past: a rule tuned to the data looks significant by construction. That is why the out-of-sample tests and the ongoing record, not this paragraph, are the real test.

How the backtest was built

Finally, the methodology. It was built to be conservative where it matters most, on price:

• Real quotes, not models: one-minute consolidated bid/ask data across the entire SPX 0DTE option chain. My research shows this is granular enough to simulate the strategies’ results with only small deviations.

• The full bid-ask spread on every leg: the short is sold at the bid and the hedge bought at the ask; a stopped trade buys back the short at the ask and sells the hedge at the bid. This is a point where many other backtests do not tell the whole truth. Slippage between bid-ask is a huge cost over long periods and thousands of trades. Calculating the full spread is as conservative as you can be in a backtest.

• Stops checked every minute against real quotes, not on 15-minute or end-of-bar data.

• Skew-aware strike selection, so the target delta is the delta the market actually prices.

• Settlement on the SPX official close, not the last print at 4:00 pm.

• $1.20 commission per leg.

• One contract per leg throughout, nothing reinvested, so the curve shows the edge, not compounding.

That last point makes the results also noticeably conservative. The margin capital has to sit in your brokerage account, but it doesn’t have to sit idle: it can be held in Treasury bills or similar. The plain backtest includes no interest on it. Over the backtest window, 3-month T-bills yielded about 4.5% on average; on a Full-size account of $443,000, that would have added roughly $203,000 over the four years, on top of the results shown. This is included in the numbers shown in The Performance page.

Feel free to explore the full statistics for each strategy and for the Portfolio. If they catch your interest, start the free trial: you will see the results since 1 May 2026, day by day for each strategy and the Portfolio, in the members’ Discord.